Abstract: Apple size is an important factor for quality classifying. In order to realize the size grading of apple, a set of apple size sorting system based on image processing was designed according to the ball shape characteristic of apple. The study selected 30 apples with the fruit diameters of 65-95 mm, simulated the multiple random status of the apple in the conveying flip device, took photos for each test sample in nine angles, and transmitted all images to the computer for pretreatment, such as binarization, denoising and edge processing, to draw the diameter of fruit surface of the apple. According to the image edge, apple’s centroid was determined, then the author sought the longest distance between the center and edge as the radius of the minimum circumscribed circles of apple image and the image in which the area between pixel and minimum circumscribed circle were closest from nine apple images, and calculated the minimum circumscribed circle diameter of surface of fruit diameter in the apple samples. Apples’ actual size was compared with the measuring size, research showed that the separation accuracy was 96.67%. The study could provide a reference for automatic sorting of apple size.